Anyone selling on Turkish e-commerce platforms knows the visual problem well. Listing a product means at least 2-3 photos. Color variants need their own sets. New campaign? New visuals. Season change? Same product, different background. Running this cycle through a traditional photo studio is expensive and slow. AI visual production breaks that cycle — but you need to know Trendyol and Hepsiburada's specific rules, which product categories respond well, and where AI runs into its limits.
What do the platforms actually require?
Both platforms require a light or white background on the main product image. Trendyol requires a minimum of 1000x1000 px. Hepsiburada wants fully white (RGB 255,255,255 or very close). These rules update regularly — since 2024, Trendyol has tightened background tolerance and rejects color deviations more often. In AI production, these requirements are straightforward: write 'clean white background, product isolated' in the prompt and it comes out right. The real challenge is consistency: 12 color variations of the same product with identical lighting, shadows, and perspective. Doing that manually takes serious time. In AI, 12 variations from one template takes the same amount of time as one.
Which product categories work best?
Textiles, accessories, home textiles, stationery, small electronics — these produce consistent results. If the product is a simple surface or silhouette, AI renders it cleanly. Where it struggles: natural textures (leather and wood sometimes produce odd outputs), transparent materials (glass bottles, clear plastic), intricate handcraft prints. For these categories, a hybrid approach works better — real photography for the hero shot, AI for background removal and variations.
How many images and which combination performs?
The first image shown in search results has the most influence on purchase decisions. The following 4-6 images matter but less so. Based on Pam Istanbul's testing across categories, the combination that consistently performs: 1 white background shot (required), 2-3 lifestyle or context images, 1 scale or detail shot, 1-2 color variation images. All of these can be AI-generated — including lifestyle, with the right reference and prompt structure.
Real numbers: a 300 SKU textile catalog
A textile client project from last autumn: traditional studio route would have required 3 days of shooting plus post-production, estimated at around 2,200 USD. With AI, the job was done in 6 hours. Cost per image: roughly 8% of the traditional route. The process: start with one real product photo, give it to the AI model as reference, generate different backgrounds and angles, then apply color-swap for variants.
When AI isn't enough
In the luxury segment, if the product's material and texture is part of the sales argument, AI usually falls short. High-end handcrafted leather bags, raw-weave textiles, hand-painted ceramics — buyers 'feel' the texture through the image. AI produces surfaces that look too perfect, too sterile. The material story doesn't come through. Simple rule: if material quality is a selling point on the product page, a real shoot is necessary. For everything else, AI is a solid option.
If you're thinking about producing your product catalog with AI, we can talk through which categories fit, what the process looks like, and what it would realistically cost. Pam Istanbul has worked on Trendyol and Hepsiburada-optimized visual production for multiple clients.